fig2

Figure 2. Schematic representation of federated learning. Federated learning is a decentralized technique for machine learning that produces an algorithm trained by utilizing collected data on decentralized devices or servers, without any exchange of the data itself. Local systems employ independent machine learning by utilizing an encrypted dataset to generate a local model. A central aggregator integrates all the local models and generates a global one. This collaborative system allows different actors to collaborate on a global machine-learning model without exchanging or uploading any data to a remote server.